comparison yet.

Fixed typo in length test.

Changed wording of error when a nonundoable transaction is attempted to be undone from "undoable transaction" to "nonundoable transaction"

Changed wording of UndoError docstring from "an attempt was made to undo an undoable transaction" to "an attempt was made to undo a non-undoable transaction"

Added Sets as distinct types within template. This will make
merging sets and mapping objects (e.g. integer sets and
integer to integer mappings) easier later.

Split template into multiple files.

Change buckets to use separate key and value arrays.
This makes set handling (null value arrays) a little simpler
and will allow us to save memory if we implement types with
small keys or values (e.g. short keys).

getting key max and min values.

Added (back) PER_USE macro to make persistence handling sane.

include files. This cleaned up the code quite a bit and should make it
pretty easy to add additional types in the future.

Fixed numerous errors with handling bucket list pointers, deletions of
subtrees, etc.

Fixed bugs in items indexing.

Moved macros for handling specific key or value types to separate
include files. This cleaned up the code quite a bit and should make it
pretty easy to add additional types in the future.

Changed BTreeItems_length so it lies. Waaa. See comment in method.

check the length after convering to a list, because btree items
have to lie about their length.

include files. This cleaned up the code quite a bit and should make it
pretty easy to add additional types in the future.

the subtransactions.

in cases where a transaction is aborted due to a storage error
(e.g. ConflictError) during two-phase commit and new objects
have been stored. The new objects were left in a state where they
incorrectly appeared to have been saved and be up to date.

Also removed unnecessary invalidation of objects in other connections
when commiting sub-transactions.

Merged in changes from Evan's branch that allows the system to
continue commiting transactions when there is a failure of only one
data manager in the last phase of two-phase commit.

inconsistent data. Added a detailed comment with explanation.

Added explanation of the semantics for calling the function passed to
the storage tpc_finish method.

important for ZEO, which may need to get tickled to get updates in
non-aynchronous mode.

This was to get a newer asynccore (or ZServers) if possible.

to database connection leaks when the mount point object was garbage
collected.

if an attempt was made to write while disconnected from the storage
server.